Problems bring along their solutions: characterization of a carbapenem-resistant Citrobacter braakii strain and its lytic phage both isolated from sewage water

Citrobacter spp. are facultative anaerobic, Gram-negative bacteria that have been known to cause infections in the urinary tract and bloodstream. These infections can be either nosocomial or community-acquired. Hospitals are considered to be reservoirs for multidrug-resistant (MDR) bacteria; as such, hospital sewage is expected to contain MDR bacteria as well. In the present study, we characterized the carbapenem-resistant Citrobacter braakii strain GW-Imi-1b1 and its lytic phage vB_CbrP_HGW_001, both isolated from sewage water. We conducted comprehensive proteogenomic, phenotypic, and physiological characterizations of GW-Imi-1b1. This included antimicrobial susceptibility testing, microscopy and motility assays. The newly isolated phage was analyzed via sequencing and electron microscopy. Its environmental stability was evaluated with respect to pH and temperature. Furthermore, phage adsorption and burst size was characterized. C. braakii GW-Imi-1b1 exhibits a broad resistance spectrum, including resistance to chloramphenicol and ertapenem. Proteomic analyses identified a plasmid-encoded GES-5 β-lactamase at constitutively high abundance independent of varying meropenem concentrations. Microscopy and motility assays revealed that GW-Imi-1b1 is a non-motile strain that produces outer membrane vesicles (OMVs). Its phage vB_CbrP_HGW_001 was identified as a Kayfunavirus within the Caudoviricetes with podovirus morphology. It displayed high host specificity, broad pH and temperature stability, an infection cycle of approximately 20 min, and a burst size of 12 PFU/cell. The observed resistance phenotype of GW-Imi-1b1 is associated with the constitutively present GES-5 β-lactamase. Due to its localization on a plasmid, this resistance mechanism may be transferable within the wastewater microbiota. The OMVs could contribute to the virulence of GW-Imi-1b1; however, further identification of their cargo is necessary. Although phage vB_CbrP_HGW_001 is characterized by a narrow host specificity and a small burst size, it proved to efficiently lyze its host. This fact and its broad stability against environmentally factors turn vB_CbrP_HGW_001 into a potential treatment option in C. braakii infections and proves phage therapy as a valuable tool to combat infections caused by MDR bacteria.

Citrobacter spp. are facultative anaerobic, Gram-negative bacteria from the order Enterobacterales. They are commonly found in various environments, including water, soil, food, and the intestinal tracts of animals and humans. However, they are generally not considered to be prevalent pathogens (Lipsky et al., 1980; Parte et al., 2020). Some species cause nosocomial and community-acquired urinary tract and bloodstream infections (Fonton et al., 2024). Less commonly, infections of the skin, intra-abdominal region, and respiratory tract occur (Samonis et al., 2009).

C. freundii is the pathogenic Citrobacter most commonly found, but other Citrobacter species such as C. koseri and C. braakii can also cause infections (Samonis et al., 2009). Especially humans with a compromised immune system and neonates are susceptible to pathogenic Citrobacter spp (Doran, 1999). For instance, a bloodstream infection caused by C. braakii was reported in an immunocompromised allogeneic stem cell transplant patient (Tollkuci and Myers, 2021). An analysis of nine newborns diagnosed with C. freundii sepsis, of which one infant died after seven hours of infection and three developed meningitis, revealed multi-drug resistances in each isolate (Chen and Ji, 2019). Antimicrobial resistance (AMR), especially multi-drug resistance, represents a major global healthcare problem in the context of infections caused by Citrobacter spp. and other pathogens (Fonton et al., 2024).

Hence, it is of utmost importance to further characterize Citrobacter pathogens and their resistance mechanisms and to come up with novel treatment strategies to fight AMR infections. The isolate C. braakii GW-Imi-1b1, analyzed in this study, was obtained from hospital sewage water in a small town in Northern Germany (Greifswald, Germany). Hospitals provide an ecological niche and select for AMR bacteria, i.e., the wastewater of hospitals contains bacteria that harbor various resistances (Hocquet et al., 2016).

C. braakii GW - Imi-1b1, hereafter GW-Imi-1b1, was detected in the inhibition zone of an imipinem-sensitive C. portucalensis isolate (Schneider et al., 2023). Imipenem is a carbapenem, a class of β-lactam antibiotics that exhibit broad-spectrum antibacterial activity against numerous pathogens, including Pseudomonas, streptococci, Neisseria and Haemophilus (Hellinger and Brewer, 1991). Carbapenems are resistant to hydrolysis by the majority of β-lactamases and are typically employed in the treatment of severe nosocomial infections caused by multi-resistant bacteria, such as extended-spectrum β-lactamases (ESBL)-producing Enterobacteriaceae and Pseudomonas aeruginosa (P. aeruginosa) (Patrier and Timsit, 2020). Therefore, it is not surprising that, according to the WHO, Enterobacterales resistant against this last-resort antibiotic currently belong to the most threatening bacteria to human health (WHO Updates List of Drug-Resistant Bacteria Most Threatening to Human Health). Their resistance is based on several mechanisms. Carbapenem-resistant Enterobacterales (CRE) often harbor carbapenemases, which are β-lactamases capable of hydrolyzing carbapenems. Non-carbapenemase resistance includes porin loss or dysfunction in combination with the production of ESBL and AmpC β-lactamases (Ma et al., 2023; Wise et al., 2024). Penicillin-binding proteins, the cellular targets of carbapenems and other β-lactam antibiotics, can be modified, thereby impeding its affinity for carbapenems and thus its antibacterial effects. Furthermore, alteration of outer membrane proteins (OMP) lead to a decreased drug uptake, whereas increased expression of efflux pumps, typically the AcrAB-TolC efflux pump, enhances carbapenem export. Lastly, biofilm remodeling has been demonstrated to induce drug resistance (Ma et al., 2023).

Consequently, limited antibiotic treatment options against multiresistant bacteria are available yet and a focus should be put on the development of alternative therapies such as phage therapy. Bacteriophages, or phages, are viruses infecting bacteria and were independently discovered by Frederick Twort in 1915 and by Félix Hubert d’Herelle in 1917 (Twort, 1915; Summers, 2017). They are characterized by a considerable degree of genomic and morphological diversity and can be classified into two categories: tailed and non-tailed virions. Furthermore, the genomes of phages can be either double-stranded or single-stranded DNA or RNA (Dion et al., 2020). However, the most prevalent phages generally contain double-stranded DNA and belong to the tailed phages. Of particular significance within the context of phage therapy against multiresistant bacteria are lytic phages (Strathdee et al., 2023). Most phages feature a narrow host range, meaning that they infect only their target host bacterium but leave other bacteria within a microbiome unaffected. Another replication strategy alongside the lytic is the lysogenic replication. Instead of host lysis, the bacteriophage genome is integrated into the bacterial genome or maintained as an episomal element. Subsequently, the viral genomic information undergoes replication within the host’s genome. However, lysogenic phages might become lytic in response to environmental changes (Kasman and Porter, 2025).

Despite the fact that phages had been discovered over a century ago, the introduction of antibiotics led to a stagnation of phage therapy research in the Western world. Given the emergence of multiresistant bacteria as a global health concern, phage therapy has resurged as a promising alternative treatment option (Gordillo Altamirano and Barr, 2019). Indeed, wastewater samples containing multidrug-resistant bacteria represent a suitable source for the isolation of specific lytic bacteriophages targeting these hosts. Numerous bacteriophages have already been successfully isolated against pathogens like Escherichia coli (E. coli), Klebsiella pneumoniae and P. aeruginosa (Alharbi and Ziadi, 2021; Fu et al., 2021; Gu et al., 2024; Kim et al., 2024).

In this study, we characterized a carbapenem-resistant C. braakii GW-Imi-1b1, which was sequenced by Schneider et al. in 2023 (Schneider et al., 2023), and a lytic phage against it from the same environment. Proteome and phenotypic analyses were employed to further characterize this strain. In addition, its lytic phage was subjected to sequencing and environmentally stability was assessed. This carbapenem resistant bacterium-phage system serves as a model system, which could be used to develop alternative treatment options against multi-resistant bacteria.

Strain information

GW-Imi-1b1 was isolated from the hospital sewage of Greifswald, Germany, in October 2020 on Klebsiella ChromoSelect Selective Agar (Merck KGaA, Darmstadt, Germany). Subsequent sequencing was conducted by Schneider et al. in 2023 (Schneider et al., 2023). The strain is deposited at the German Collection of Microorganisms and Cell Cultures (DSMZ) as DSM 120033. Strains used for investigation of lytic host spectrum were obtained from the strain collection of the laboratory of Katharina Schaufler (Helmholtz Institute for One Health, Germany). The use of bacterial isolates was approved by the Ethics Committee of the Christian-Albrechts-University Kiel, Germany (approval no. D 583/22). All samples were obtained as part of routine diagnostics and were fully anonymized prior to analysis; therefore, informed consent was not required. The strains C. braakii (DSM 17596) and C. freundii (DSM 30039) were obtained from the DSMZ (Braunschweig, Germany).

Phenotypical characterization of C. braakii GW-Imi-1b1

Scanning electron microscopy

To prepare the cells for scanning electron microscopy, 1 mL of a mid-exponential GW-Imi-1b1 culture in lysogeny broth (LB) medium was mixed with 9 mL of 0.9 % NaCl. The suspension was then filtered through a polycarbonate filter (pore size 0.2 µm; Merck Millipore, Darmstadt, Germany). A piece of filter (approximately 1 sqcm) was transferred to 1 mL of a fixative solution (20 mM HEPES, 1 % glutaraldehyde, 4 % paraformaldehyde, 0.2 % picric acid), incubated for 1 h at room temperature (RT) and then stored overnight at 4 °C. Samples were washed with washing buffer [100 mM cacodylate buffer (pH 7), 1 mM CaCl 2 ] three times for 10 min each time, treated with 2 % tannic acid in washing buffer for 1 h at RT, and washed again with washing buffer three times for 15 min each time. After that, samples were dehydrated in a graded series of aqueous ethanol solutions (10 %, 30 %, 50 %, 70 %, 90 %, and 100 %) on ice for 15 min each time. Before the final change of 100 % ethanol, samples were allowed to reach RT and then critical point-dried with liquid CO 2. Finally, samples were mounted on aluminum stubs, sputtered with gold/palladium and examined with a field emission scanning electron microscope Supra 40 VP (Carl Zeiss Microscopy Deutschland GmbH, Oberkochen, Germany) using the Everhart-Thornley SE detector and the in-lens detector in an 80:20 ratio at an acceleration voltage of 5 kV. All micrographs were edited using Adobe Photoshop CS6.

Transmission electron microscopy

To visualize the cellular flagellation of cells as well as the purified phages, the flotation method was used for the negative staining procedure.

To investigate the cellular flagellation, 3 mL of a GW-Imi-1b1 cell culture in the mid-exponential phase were centrifuged. Cells were washed in PBS and then fixed with a solution of 1 % glutaraldehyde and 4 % paraformaldehyde in PBS for 1 h at RT and then stored at 4 °C overnight. After that, the cells were washed twice with PBS and finally resuspended in 100 µl PBS. Fixed cells were allowed to adsorb onto a glow-discharged Pioloform carbon-coated 400-mesh grid for 10 min. The grid was then transferred onto two droplets of deionized water, and finally onto a drop of 0.5 % aqueous uranyl acetate for 10 s. After blotting with filter paper, the grids were air-dried.

For preparing phages for TEM, 1 mL of a high titer phage suspension was centrifuged at 25,000 x g for 60 min. After washing of the resulting pellet with 1 mL 0.1 M ammonium acetate (pH 7), phages were dissolved in 300 µl Preissner buffer (4 mM KH 2 PO 4, 6 mM K 2 HPO 4). Afterwards, phages were allowed to adsorb onto a glow discharged carbon-coated holey Pioloform film on a 400-mesh grid for 5 min. The grid was then transferred onto two droplets of deionized water, and finally onto a drop of 1 % aqueous uranyl acetate for 15 s. After blotting with filter paper, the grids were air-dried.

All samples were examined with a transmission electron microscope LEO 906 (Carl Zeiss Microscopy Deutschland GmbH, Oberkochen, Germany) at an acceleration voltage of 80 kV. For acquisition of the images, a wide-angle dualspeed CCD camera Sharpeye (Tröndle, Moorenweis, Germany) was used, operated by the ImageSP software. Finally, all micrographs were edited by using Adobe Photoshop CS6 or Affinity Photo 2.

Resistance profile

Antimicrobial susceptibility testing of GW-Imi-1b1 was performed using the disc diffusion method, also known as the Kirby-Bauer method, according the EUCAST guidelines (The European Committee on Antimicrobial Susceptibility Testing, 2026). Briefly, the strain was cultured in Mueller-Hinton medium (MH; Carl Roth GmbH, Karlsruhe, Germany) overnight and cell density was adjusted to a 0.5 McFarland turbidity standard. The cell suspension was spread on MH agar plates using sterile cotton swabs. Antimicrobial discs (bestbion dx GmbH, Hürth, Germany; see details below) were placed firmly on the surface of the inoculated agar plate. Agar plates were incubated upside-down at 37 °C for 24 h. Subsequently, inhibition zone diameters were measured and interpreted according to the EUCAST breakpoints. The following antibiotic discs were used: amikacin (AK) 30 µg, amoxicillin + clavulanic acid (AUG) 30 µg, aztreonam (ATM) 30 µg, cefotaxime (CTX) 5 µg, ceftazidime (CAZ) 10 µg, chloramphenicol (C) 30 µg, ciprofloxacin (CIP) 5 µg, ertapenem (ETP) 10 µg, imipenem (IMI) 10 µg, levofloxacin (LEV) 5 µg, meropenem (MRP) 10 µg, ofloxacin (OFX) 5 µg, tobramycin (TOB) 10 µg, trimethoprim/sulfamethoxazole (SXT) 25 µg, cefepime (CEP) 30 µg.

Epsilometer tests (E-test) to determine the minimal inhibitory concentration (MIC) of carbapenem antibiotics were done in the same fashion as disc assays on MH agar plates using MIC test strips for ertapenem, imipenem, and meropenem, with a concentration ranging from 0.002-32 µg/mL (bestbion dx GmbH, Hürth, Germany). The results were evaluated according to the breakpoint table in the EUCAST guidelines, version 16.0, 2026 (The European Committee on Antimicrobial Susceptibility Testing, 2026).

Validation of the present GES-5 β-lactamase was performed by plasmid isolation of an overnight culture of GW-Imi-1b1 with Monarch ® Plasmid Miniprep Kit according to the manufacturer’s protocol (New England Biolabs), followed by a PCR to amplify the expected locus of bla GES-5.

Characterization of virulence factors

To screen for potential virulence factors of GW-Imi-1b1, motility assays and a protease secretion assay were conducted. All assays were performed with P. aeruginosa PAO1 as the positive control (Holloway and Morgan, 1986), as this strain is known to secrete proteases and exhibit different motilities (Filloux, 2011; Kazmierczak et al., 2015; Jose and Singh, 2020).

Motility assays included swimming, swarming, and twitching motility tests. The different media were prepared according to the protocol outlined by Rashid and Kornberg (2000). For swimming and swarming motility, overnight cultures were prepared by inoculating 5 mL LB and incubating overnight at 37 °C at 150 rpm. To assess the swimming motility, 5 µl of cell culture was positioned at the center of the specific agar (10 g/l trypton, 0.3 % agar, 5 g/l NaCl) and subsequently incubated at 37 °C for 16 h. Swarming motility was tested just as swimming motility, but using a different semi-solid medium (8 g/l LB, 0.5 % glucose, 0.5 % agar) and subsequent incubation at 37 °C for 16 h. A modified protocol based on (O’Hara et al., 2024) was used to examine twitching motility (O’Hara et al., 2024). In brief, a colony from the respective strain was pricked through the agar (10 g/l trypton, 5 g/l yeast extract, 10 g/l NaCl, 1 % agar) and incubated for 48 h at 37 °C. Subsequently, the agar was detached from the plate and the petri dish was stained with a 0.1 % crystal violet solution for a period of 10 min at RT. Following a gentle wash with A. dest., a twitching zone was observed on the surface of the petri dish. Motility was documented in pictures after the respective incubation time and the areas determined using the NIH ImageJ software 1.54g (Schneider et al., 2012). Additionally, the genome of GW-Imi-1b1 (GenBank: CP115723.1) was analyzed in silico to identify putative involved genes associated with bacterial motility and adhesion systems. Therefore, we screened the corresponding protein sequences against the Pfam-A database (Pfam 38.2) using hmmsearch implemented in HMMER v3.4 (Eddy and the hmmer development team, 2023). Pfam domain identifiers (Pfam-ID) corresponding to known flagellar, Type IV pili and Type I fimbriae proteins were selected based on their functional annotations in the Pfam database. Pfam domain hits were considered significant using an E-value cutoff of 1e-5.

Protease secretion was examined in different growth phases. For that, LB medium was inoculated to a OD 540nm of 0.08 from a previously prepared 5 mL LB overnight culture, and the bacteria incubated at 37 °C and 150 rpm. Cells were harvested at the exponential (t1: OD 540nm 0.5), transient (t2: OD 540nm 1), stationary (t3: OD 540nm 2), and late stationary phase (t4: OD 540nm:3-4). The culture was sterile filtered through a 0.2 µm filter and subsequently used for protease activity assays. P. aeruginosa PAO1 was used as positive control, the corresponding quorum-sensing-deficient mutant P. aeruginosa Δ lasI/rhlI (Hentzer et al., 2003) was used as negative control. Protease secretion was tested by the cleavage of azocasein. In the presence of proteases, the azo dye is cleaved from casein. Consequently, the released dye is quantifiable through photometric analysis. Cleavage can be done by several proteases like serine proteases, metalloproteases and acid proteases (Peyronel and Cantera, 1995). The detection of protease secretion was assessed by a modified protocol after Hentzer et al. (2002). To this end, an azocasein solution was prepared composed of 50 mM Tris, 5 mM EDTA and 1 % azocasein and adjusted to a pH of 7.5. Some 150 µl of the respective sterile filtered culture medium were mixed with 250 µl of the azocasein solution and incubated at 37 °C for 6 h. To this, some 1.2 mL of 10 % TCA were added and again incubated for 25 min at RT. Following centrifugation for 15 min at 21,382, 600 µl of the sample were mixed with 750 µl of 1 M NaOH in a cuvette. Absorption was subsequently measured at 440 nm. For reference, a negative sample consisting of 250 µl of azocasein solution, 150 µl of LB and 1.2 mL of 10 % TCA were mixed and 600 µl of the mixture was added to 750 µl of 1 M NaOH. The protease activity is proportional to the intensity of the dye measured at an extinction of 440 nm and divided by OD 540nm of the respective time point of harvest. Statistical significance was assessed using an unpaired two-sample t-test (Welch’s correction) in RStudio 1.4.33.

Proteome analysis of C. braakii GW-Imi-1b1

To characterize the protein profile of GW-Imi-1b1, extracellular and cellular proteins were extracted and analyzed. For that, the strain was cultivated in LB to early stationary phase (OD 540nm 1.5). Subsequently, 30 mL of the culture were harvested and centrifuged for 10 min at 4 °C and 9,693 x g. For cellular protein extraction, the resulting pellet was washed twice in 1 mL TE-Buffer (10 mM Tris, 1 mM EDTA, pH 7.5) and subsequently resuspended in 1 mL TE-Buffer with 1 % Triton-X. The suspension was transferred into a 2 mL screw cap micro tube with 500 mg of glass beads (0.1 mm diameter) and mechanically disrupted three times with a FastPrep-24TM5G (MP Biomedicals, Irvine, USA) at 6.5 m/s for 30 sec. After centrifugation for 5 min at 4 °C and 21,382 x g, the supernatant was transferred into a new reaction tube and centrifuged again to remove residual glass beads and cell debris. Subsequently, six volumes of -20 °C cold acetone was added, and the resulting solution was incubated overnight at -20 °C. The next day, the sample was centrifuged again for 1 h at 4 °C and 21,382 x g and the supernatant removed. The precipitated proteins were washed in 1 mL acetone and subsequently dried in a vacuum centrifuge. The dried pellet was resuspended in 150 µl of 8 M urea and 2 M thiourea with shaking at 150 rpm. Samples were stored at -20 °C until further processing.

Extracellular proteins were precipitated from the supernatant after harvesting with TCA. 3.3 mL 100 % TCA were added to the 30 mL of supernatant and incubated overnight at 4 °C. Samples were then centrifuged for 1 h at 4 °C and 9,693 x g. The resulting protein pellet was dissolved in 1 mL of 70 % ethanol and again centrifuged for 5 min at RT and 16,060 x g in a new reaction tube. This washing step was repeated once with 70 % ethanol and 100 % ethanol. Subsequently, protein pellets were dried in a vacuum centrifuge and resuspended in 50 mL 8 M urea and 2 M thiourea. Samples were stored at -20 °C until further processing. Protein concentration was determined with ROTI ® Nanoquant (Carl Roth GmbH, Karlsruhe, Germany) according to the manufacturer’s protocol.

Effect of meropenem on the protein profile of C. braakii GW-Imi-1b1

For pre-culturing, we inoculated 100-mL Mueller Hinton Broth (MHB) in 500-mL Erlenmeyer flasks with one colony of a 24 h culture (37 °C, MH agar plates) of GW-Imi-1b1 and incubated overnight (37 °C, 180 rpm). The meropenem-sensitive strain C. portucalensis St35 was used as a positive control. Material from these cultures was used to inoculate 100-mL fresh MHB to a final OD of 0.1, and meropenem stock solution (5 mg/mL in PBS) was added to all but the control flasks to reach final concentrations of 0.064, 0.125, or 20 µg/mL meropenem, respectively. Concentrations were chosen to cover the range indicated by the E-test, including subinhibitory levels. Cultures were incubated at 37 °C and 180 rpm, and samples for proteomics analysis were taken from the overnight-cultures at the start of the experiment (t0), as well as from flasks after 2 h (t1) and 24 h (t2) incubation. Samples (4.5 mL each) were washed once with PBS (10,000 x g, 4 °C) and the pellets stored at -70 °C until further processing. Additionally, bacterial growth was measured by determining the optical density OD 500nm of the culture. Experiments were performed in biological triplicates.

Proteins were extracted from GW-Imi-1b1 as previously described (Hinzke and Markert, 2017) with minor modifications. In brief, lysis buffer (50 mM TEAB, 4 % SDS, 50 mM DTT) was added in a 1:5 ratio to pellets, the pellets were carefully resuspended, samples heated (10 min, 95 °C) and subsequently sonicated in an ultrasonic bath for 10 min. Samples were cooled on ice in between. Protein concentration was determined with ROTI ® Nanoquant (Carl Roth GmbH, Karlsruhe, Germany) according to the manufacturer’s protocol.

Protein digestion, mass spectrometry analysis and identification

Digestion of proteins using S-trap micro columns was done as described by Brauer et al. (Brauer et al., 2022). Briefly, 50 µg of protein sample were adjusted to 5 % SDS, proteins reduced with 10 mM DTT, alkylated with 20 mM IAA and samples acidified with 12 % phosphoric acid. Proteins were then bound to the S-Trap column and digested with pre-activated trypsin for 3 h at 47 °C. Peptides were eluted in three steps with 50 mM TEAB, with 0.1 % acetic acid and with 60 % acetonitrile (ACN) with 0.1 % acetic acid. Elutions of the same sample were pooled, dried in a vacuum centrifuge and stored at -70 °C until further processing. The samples were fractionated with Pierce C18 spin columns (ThermoFisher Scientific, Waltham, USA) and application of 18 mg C18 Reprosil Gold (Dr. Maisch HPLC GmbH, Ammerbruch-Entringen, Germany) in 8 fractions using increasing amounts of ACN (5 to 50 % ACN) in 0.1 % triethylamine (Carl Roth GmbH; Karlsruhe, Germany). Subsequently, fractions 1 and 5, 2 and 6, 3 and 7 and 4 and 8 were combined to generate four final fractions. Fractionated samples were transferred to glass vials, dried in a vacuum centrifuge and stored at -80 °C until analysis.

For MS analysis, fractionated samples were resuspended in 10 µl 0.1 % acetic acid and analyzed using an EASY-nLC1200 coupled to a QExactive HF mass spectrometer (Thermo Fisher Scientific, Waltham, USA). Peptides were separated on a self-packed analytical column (100 µm inner diameter, reverse phase C18 material (Dr. Maisch HPLC GmbH, Ammerbruch-Entringen, Germany; 3.6 µm particle size, 20 cm column length) using a 80 min binary non-linear gradient (4 % to 50 % of 0.1 % (v/v) acetic acid in ACN) at a flow rate of 300 nl/min. Survey scan resolution was 60,000 in a range of 333–1650 m/z and the 15 most-intense peaks per scan were selected for fragmentation. Ions with single charge and ions with unknown charge state were excluded from fragmentation; and precursor ions were dynamically excluded for 30 s from fragmentation. Lock mass correction was enabled (lock mass 445.12003 Da).

Protein identification and label-free quantification was done using MaxQuant (v1.6.17.0) using a protein sequence database of GW-Imi-1b1 (5, 338 entries); reverse sequences and common laboratory contaminants were added by the software (Tyanova et al., 2016a). Protein sequences were generated from the assembled genome and were clustered at 99 % identity using CD-Hit (Li and Godzik, 2006). The following parameters were set: maximum of 2 missed cleavages, methionine oxidation as variable modification, carbamidomethylation of cysteine as fixed modification, at least one unique peptide required, match between runs enabled. Unique and razor peptides were considered for label-free quantification with a minimum ratio count of 2. For further analysis, we used Perseus v2.1.3.0 for protein quantification analysis and statistical evaluation, pSORTBv3.0.2 for prediction of subcellular localization, and eggNOG-mapper version 2.1.9 for functional annotation of the identified proteins (Yu et al., 2010; Tyanova et al., 2016b; Cantalapiedra et al., 2021).

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with dataset identifier PXD063554 (Perez-Riverol et al., 2024).

Lytic phage vB_CbrP_HGW_001

Phage isolation from sewage water

For isolation of lytic phages against GW-Imi-1b1, 3 mL of wastewater of the inflow to the sewage treatment plant of Greifswald in November 2023 was mixed with 30 µl chloroform to eliminate present bacteria. Some 200 µl of the treated water was then incubated with 200 µl of an overnight culture of GW-Imi-1b1 for 10 min at RT. Phage presence was tested using the double agar layer method according to Kropinski et al. (2009). Briefly, 200 µl of the incubated culture was added to soft agar (0.75 % agar in LB medium) pre-tempered to 45 °C. After short mixing, the soft agar mix was poured onto base agar containing 1.5 % agar and solidified at RT. The plate was subsequently incubated at 37 °C overnight.

On the next day, visible plaques were isolated by pricking the top agar layer with a pipette tip. Next, the sample was suspended in 500 µl SM-buffer (0.1 M NaCl, 2.3 mM MgSO 4) and vortexed for 30 sec. After centrifugation at 14,000 x g at 4 °C for 2 min, the solution was filtered through a 0.2 µm cellulose acetate filter (VWR International). The final phage suspension was tested again for phage activity with the double agar layer method. This procedure was repeated three times to ensure phage purity. For sequencing, DNA was extracted using Monarch ® Mag Viral DNA/RNA Extraction Kit (New England Biolabs). Two mL of a high titer phage suspension (63, 3*10 8 PFU/mL) was filtered using Amicon ® Ultra 0.5 mL centrifugal filters MWCO 3 kDa (Sigma-Aldrich) for phage enrichment. The resulting 200 µl filtrate were used for DNA extraction following the manufacturer’s protocol. DNA concentration was fluorometrically quantified by Qubit 3.0 Fluorometer (Life Technologies, Carlsbad, California, US) and adjusted to a final concentration of 50 ng/µl in 30 µl. Subsequent DNA sequencing was conducted by Eurofins Genomics via Oxford Nanopore Sequencing as part of the Whole Plasmid Sequencing Service. The genomic organization, whether it is a circular or linear dsDNA phage, was determined by subsequent restriction digestion with BamHI, EcoRI, and SapI (New England Biolabs) according to the manufacturer’s protocol. We selected the restriction enzymes to generate distinct fragment patterns, allowing determination of the phage’s circular or linear genome structure. The analysis of the digested samples was performed on a 0.4 % agarose gel.

Annotation and functional prediction of vB_CbrP_HGW_001 genes

The genome of phage vB_CbrP_HGW_001 was annotated with Pharokka v1.8.2 (Bouras et al., 2023). Specifically, coding sequences (CDS) were predicted with PHANOTATE (McNair et al., 2019), tRNAs were predicted with tRNAscan-SE 2.0 (Chan et al., 2021), tmRNAs were predicted with Aragorn (Laslett and Canback, 2004) and CRISPRs were predicted with CRT (Bland et al., 2007). Functional annotation was generated by matching each CDS to the PHROGs (Terzian et al., 2021), VFDB (Chen et al., 2005) and CARD (Alcock et al., 2020) databases using MMseqs (Steinegger and Söding, 2017) and PyHMMER (Larralde and Zeller, 2023). Contigs were matched to their closest hit in the INPHARED database (Cook et al., 2021) using mash (Ondov et al., 2016). The resulting Genbank output file was subsequently updated with functional predictions and further visualized using a Circos plot via phold (Bouras et al., 2026).

Host spectrum

The lytic host spectrum of vB_CbrP_HGW_001 was determined using Citrobacter species and other Enterobacteriaceae species (Table 1). Strains were tested using the double agar layer method, adding 10 µl of the phage suspension on the respective plate. After incubation at 37 °C overnight, lysis was determined as clearance in the phage suspension zone.

Phage adsorption assay

Phage adsorption was tested as described by Wittmann et al. with minor modifications (Wittmann et al., 2014). Briefly, exponentially growing cultures of GW-Imi-1b1 were infected with vB_CbrP_HGW_001 with a multiplicity of infection (MOI) of 0.01. The infected culture was then incubated at 37 °C and 1 mL samples were taken at 5 min intervals for a duration of 30 min. Samples were centrifuged at 4,656 x g for 4 min. The titer of free phages was subsequently determined with double agar layer method. After incubation overnight at 37 °C, resulting plaques were enumerated and the plaque forming units (PFU) were calculated. The adsorption assay was conducted in three biological replicates.

Burst size

Burst size experiments were performed as previously described by Liao et al. with slight modifications (Liao et al., 2022). Some 30 mL of a mid-exponential phase of GW-Imi-1b1 were infected with vB_CbrP_HGW_001 at a MOI of 0.01 and incubated for 5 min at 37 °C to ensure phage attachment to host cells. The culture was then centrifuged for 3 min at 6, 204 x g and washed twice with 2 mL LB. The resulting pellet was resuspended in 30 mL LB and incubated at 37 °C for 30 min. Aliquots of the culture were harvested every 5 min and filtered through a 0.2 µm polycarbonate filter. Phage titers were determined at each time point using the double-layer agar method. Burst size was calculated as the ratio of the phage titer during the plateau phase (average of t15, t20 and t25), representing complete phage release, to the number of adsorbed phages, defined as the initial phage titer after subtraction of the non-adsorbed phages in the early time points (average of t0-t5).

Temperature and pH stability

Temperature and pH stability were determined according to Akhwale et al. with minor modifications (Akhwale et al., 2019). For phage temperature stability analysis, a phage suspension of 2.15x10 10 PFU/mL was subjected to various temperatures (21 °C, 30 °C, 37 °C, 40 °C, 50 °C and 60 °C) for 6 h. Thereafter, the samples were cooled on ice, and the titer was determined by the double agar layer method. For testing of the pH stability, 100 µl of the phage suspension of 2x10 10 PFU/mL was incubated in 900 µl LB with different pH values (2, 4, 6, 7.2, 8, 10) and subsequently incubated for 6 h at RT. Phage titer was again determined by the double agar layer method directly after incubation time. A temperature of 4 °C and a pH value of 7.2 served as the controls, since the phage suspensions were stored under these conditions. Statistical significance of the results compared to the control conditions was assessed using an unpaired two-sample Welch’s t-test in RStudio version 4.4.3 (R Core Team, 2025).

Phenotypical analyses of C. braakii GW-Imi-1b1

Morphology

The scanning electron micrograph of GW-Imi-1b1 (Figure 1) revealed rod-shaped bacteria that measured approximately 2 µm in length and 0.5 µm in width, which is consistent with other members of the Enterobacteriaceae family (Morales-López et al., 2019). Interestingly, they apparently produce outer membrane vesicles (OMVs).

Protease secretion

GW-Imi-1b1 secreted no proteases, as shown by azocasein cleavage comparable to that of the negative control strain P. aeruginosa PAO1 ΔlasI/rhlI. Protease secretion was only detected the positive control strain P. aeruginosa PAO1 (Figure 2).

Motility of C. braakii GW-Imi-1b1

In silico analysis of the GW-Imi-1b1 genome identified 92 putative motility-related domains, including 34 associated with flagellar structure, 8 with Type IV pili and 49 with Type I fimbriae (see Supplementary Table S3). However, whole-proteome analysis detected only ten proteins of the aforementioned functional categories. These included four flagellar proteins, five Type I fimbrial proteins and one Type IV pilus protein (see Supplementary Table S3). In agreement with these findings, no motility phenotype was observed in motility assays, as growth remained confined to the inoculation spot in all assays. These results indicate that GW-Imi-1b1 appears to be incapable of swarming, swimming or twitching motility (Figure 3). In contrast, the positive control strain P. aeruginosa PAO1 exhibited clear motility under each condition. This motility included a characteristic swarming pattern, as evidenced by the formation of tendrils or dendrites (Jose and Singh, 2020), observable swimming-dependent movement through semiliquid medium, and twitching motility by pilus-mediated surface movement. Furthermore, the transmission electron microscopy of GW-Imi-1b1 (see Supplementary Figure 3) indicated that the cells do not produce flagella, supporting the observed lack of swimming and swarming motility.

Antibiotic resistance profile of C. braakii GW-Imi-1b1

GW-Imi-1b1 exhibits resistance to multiple antibiotics (Table 2). While GW-Imi-1b1 was resistant to tobramycin, the strain remained susceptible to the other aminoglycosides tested, e. g. amikacin. Cefotaxime showed activity only at increased exposure, while the other cephalosporin, ceftazidime, exhibited no inhibitory effect. Likewise, the strain was resistant to the fluoroquinolones ciprofloxacin, levofloxacin, and ofloxacin. GW-Imi-1b1 was only sensitive to carbapenems imipenem and meropenem at high concentrations, and resistant to ertapenem, with an average MIC of 12 µg/mL. According to the EUCAST guidelines, isolates were only susceptible at higher exposures for meropenem >2 µg/mL and imipenem >2 µg/mL (The European Committee on Antimicrobial Susceptibility Testing, 2026). The Supplementary Figure 2 contains images of E-tests and disk diffusion results. In addition to laboratory analyses, we identified genes mediating antimicrobial resistance in the bacterial genome and plasmids in silico using Resfinder v4.7.2 (Camacho et al., 2009; Bortolaia et al., 2020) (see Supplementary Table S3). We found several AMR-associated genes located on both plasmids and the chromosome, conferring resistance to β-lactams, chloramphenicol, fluoroquinolones, aminoglycosides and sulfonamides. The strain was predicted to be resistant to sulfamethoxazole, which was disproved in the conducted resistance tests.

Proteome analysis of C. braakii GW-Imi-1b1

We detected 2, 345 proteins in the whole cell fraction and 1, 291 proteins in the extracellular fraction in at least two out of four replicates. Subsequent subcellular prediction by pSORTB v3.0.2 excluded identified cytoplasmic proteins, resulting in a total of 352 non-cytoplasmic proteins being identified (Yu et al., 2010). The high contamination rate of the proteome with cytosolic proteins is probably a result of increased cell lysis at the time of point of harvest (early stationary phase). We further analyzed the potential function of identified proteins by eggNOG-mapper version 2.1.9 (Cantalapiedra et al., 2021) (Figure 4). In Table 3, selected identified proteins associated with secretion systems and antibiotic resistance are listed. With regard to the resistance-associated proteins, the following were identified: several β-lactamases, components of efflux pumps and transporters, and antibiotic-modifying enzymes were expressed by GW-Imi-1b1. The β-lactamase (Protein ID: 05428 GenBank: WBU75835.1) is among the ten most abundant proteins in the whole cell proteome, despite the absence of antibiotic stress during bacterial growth. We found the corresponding gene to be present on plasmid CP115731. It was further identified as a carbapenem-hydrolyzing class A β-lactamase Guiana Extended-Spectrum (GES)-5 (Schneider et al., 2023). The GES-5 gene is known to be wide distributed across various bacterial species (Ellington et al., 2019). Plasmid isolation and subsequent PCR of bla GES-5 confirmed its localization on CP115731 (Supplementary Figure 4). All identified proteins, including EggNOG analysis results are provided in the Supplementary Table S3.

Effect of meropenem on C. braakii GW-Imi-1b1

As described above, the proteome profile of GW-Imi-1b1 exhibited an unusually high abundance of a GES-5 β-lactamase. The subsequent growth comparison of GW-Imi-1b1 and C. portucalensis St35 with different meropenem concentrations exhibited a marked difference (Figure 5). While the growth of GW-Imi-1b1 was almost not affected at all, already low concentrations of meropenem led to growth inhibition of C. portucalensis St35. Along incubation time, samples for mass spectrometric analysis were harvested from the pre culture (t0), after 2 h of growth (t1) and after 24 h (t2).

The proteome was analyzed with a focus on the abundance of the GES-5 β-lactamase (ID: 05428). Independent of the presence or concentration of meropenem the specific β-lactamase was among the most abundant proteins in GW-Imi-1b1 (Figure 6). ANOVA with a FDR threshold of 0.05 across all conditions revealed no significant differences in the abundance of this protein (-log 10 p = 1.268). For C. portucalensis St35, no GES-5 β-lactamase was detected but other β-lactamases in lower abundances (PRIDE dataset indentifier: PXD063554).

The lytic phage vB_CbrP_HGW_001

The successfully isolated bacteriophage formed plaques measuring 3–4 mm in diameter and were surrounded by a halo (see Supplementary Figure 5A). The complete genome sequence of vB_CbrP_HGW_001 comprises 39,389 base pairs (bp) with a total GC Content of 50.04%. Notably, its host strain GW-Imi-1b1 exhibits a similar GC Content of 51.81 % (Schneider et al., 2023). In silico analyses of the phage genome revealed 57 potential coding sequences (CDS) (Figure 7B). No genomic features like tRNAs, CRISPRs, tmRNAs were identified. Furthermore, no coding regions regarding integration and excision were found, indicating an exclusively lytic phage. However, 26 of the 57 CDS were annotated as having unknown functions (see Supplementary Table S1). The phage vB_CbrP_HGW_001 has a short, non-contractile tail and an icosahedral capsid as revealed by transmission electron microscopy (Figure 7A). The phage particles have a head diameter of 54 ± 3 nm and short tails of 11, 8 ± 2, 3 nm. Consequently, we taxonomically classify it as belonging to the Caudoviricetes. Furthermore, the genome of vB_CbrP_HGW_001 aligns with known Kayfunavirus phage genomes (see Supplementary Table S2). Based on this analysis and the restriction digestion with BamHI, EcoRI, and SapI (see Supplementary Figure 5B), vB_CbrP_HGW_001 is likely a circular dsDNA bacteriophage.

The lytic host spectrum was restricted to the propagation strain GW-Imi-1b1, as no lysis was detected in any of the other tested strains (Table 1).

The replication of the phage, along with the subsequent lysis of the host cell, occurred within a period of 20 min, as indicated by the increase in the titer of non-adsorbed phages after this time duration (Figure 8A). Almost immediately after infection, the phages were adsorbed, as nearly no phage particles were detectable after 5 min. As demonstrated in Figure 8B, the one-step growth curve validated the brief duration of the lytic replication cycle. This is evidenced by its approximately 5-min latent period, followed by complete phage release within the subsequent 10 min, resulting in a plateau phase reached between 15 and 25 min. The resulting burst size was computed as 12 virus particles per cell.

Stability of phage suspension was tested by incubation at various temperatures and pH values. Figure 8C revealed a significant decrease of PFUs at 50 °C and 60 °C. These findings indicate the phage retains stability over a temperature range of 4 °C to 40 °C. The test further indicated that a pH value ranging from 6 to 10 does not exert an effect on phage stability (Figure 8D). Nonetheless, a decline in the number of infectious phage particles was observed at increasingly acidic pH values with no infectious particles left at a pH value of 2.

In this study, we characterized a carbapenem-resistant C. braakii strain and its lytic bacteriophage, both isolated from sewage. There is a high potential for the production of OMVs of GW-Imi-1b1, as evidenced by electron microscopy. OMVs were also reported in other Gram-negative pathogens such as Helicobacter pylori, P. aeruginosa, and Citrobacter rodentium (Jan, 2017; Sinha et al., 2019). They are further associated with a variety of inflammatory diseases. As OMVs comprise lipopolysaccharides, peptidoglycans, and additional virulence factors, including enzymes and toxins, they are capable of triggering an immune response. However, in addition to their pathogen-associated function, OMVs have been attributed to inter-bacterial interactions. Thus, OMVs have the potential to carry antibiotic resistance genes that can be transferred to susceptible cells (Chen et al., 2023; Zhao et al., 2025). Therefore, GW-Imi-1b1 may use its OMVs for both inter-bacterial interactions and/or virulence. Isolation and subsequent identification of the cargo is necessary to unravel the role of OMV production in Citrobacter braakii. To evaluate the importance of the OMVs in terms of horizontal gene transfer, DNA isolated from their cargo could be sequenced and subsequently aligned with the whole-genome sequence of GW-Imi-1b1 in future studies. This approach could enable the detection of putatively transferred genes, including antimicrobial resistance determinants. Most notably, OMVs of other Gram-negative bacteria have been described to facilitate resistance to phage attack. By acting as decoy, OMVs could bind phage particles and thus protect the bacterial cell from phage binding (Reyes-Robles et al., 2018; Augustyniak et al., 2022; Xuan et al., 2024). Hence, the observed strong potential of OMV synthesis by GW-Imi-1b1 could also serve as a passive phage resistance mechanism.

GW-Imi-1b1 does, based on our findings, not secrete proteases. Although GW-Imi-1b1 expresses both the Tat- and Sec-secretion system, no protease secretion was detected under the conditions tested. Moreover, the mass spectrometric analysis identified only cytoplasmatic and periplasmatic proteases, a finding that aligns with the results of extracellular protease testing. In contrast, serine proteases, which are also known as serine protease autotransporters of Enterobacteriaceae (SPATE), are secreted by several members of this family, including species of Shigella, Salmonella and Citrobacter (Navarro-Garcia, 2023). However, despite the reported occurrence of SPATEs in Citrobacter, such as Citrobacter rodentium (Yen et al., 2008), no protease activity was detected in other members of the genus, including the opportunistic pathogen C. freundii (Chen et al., 2024). Consequently, the absence of protease secretion may not directly impact virulence, although it has been reported that proteases contribute to the virulence of pathogens by degrading host cell membrane proteins and damaging tissues (Marquart, 2021).

Motility assays demonstrated that GW-Imi-1b1 is completely immotile under tested conditions with respect to swimming, swarming and twitching. Furthermore, as to our knowledge no motility has been reported for C. braakii strains so far. Swimming motility has been reported for the genus Citrobacter, for example in Citrobacter werkmanii (Zhou et al., 2023). However, motility might be induced under certain circumstances, as C. freundii has been shown to exhibit swarming motility in the late stationary phase on 0.5-0.7% agar (Cong et al., 2011). Transmission electron microscopy and the low presence of motility-related proteins detected in proteome analysis support the hypothesis of an absence of flagella. Flagella are especially needed for the purpose of swimming and swarming motility (Wadhwa and Berg, 2022). As a flagellum is usually composed of about 30 different proteins ranging from a few to tens of thousands copies (Minamino and Kinoshita, 2023), the presence of a fully assembled flagellum in GW-Imi-1b1 seems unlikely. The absence of motility may thus affect the virulence of this strain, given that flagella are known to contribute to the establishment of bacterial infections (Xia et al., 2026).

We found a GES-5 β-lactamase (Protein-ID:05428) in strikingly high abundance comparable with that of some ribosomal proteins and independently from antibiotic presence. Moreover, sequencing by Schneider et al. of GW-Imi-1b1 located the gene encoding for this β-lactamase on a plasmid (Schneider et al., 2023). Further PCR in this study confirmed the gene locus. Hence, it could potentially be transferred by conjugation or transformation within the wastewater microbiota, thus transferring resistance to other organisms. Case reports have already documented a plasmid-encoded GES-5 β-lactamase and its transmission in different species, including E. coli, Klebsiella oxytoca and Enterobacter cloacae, across various patients and hospital settings (Ellington et al., 2019). Furthermore, conjugation with a plasmid harboring bla GES-5 in Pseudomonas knackmussii strains resulted in an increased resistance to several carbapenems (Girlich et al., 2012). Given the observed carbapenem resistance and the unimpeded growth of GW-Imi-1b1 under varying meropenem concentrations, it is plausible that this β-lactamase also contributes to the resistance phenotype. Furthermore, a proteome analysis of the carbapenem-sensitive strain C. portucalensis St35 revealed similar proteins involved in antibiotic resistance, but lacking the aforementioned β-lactamase (PRIDE dataset indentifier: PXD063554). The sensitivity of C. portucalensis St35 to meropenem underlines this hypothesis. Several other antibiotic resistances of GW-Imi-1b1 were observed, in agreement with the results of the proteome analysis. The AcrAB-TolC efflux system, of which we detected all subunits in the proteome, contributes to resistance to multiple antibiotics by actively exporting them (Nishino et al., 2021). These findings suggest that, in addition to the GES-5 β-lactamase, other resistance mechanisms could thus contribute to the carbapenem resistance phenotype.

Taken together, GW-Imi-1b1 produces various virulence factors, including secretion systems, efflux pumps and OMVs as evidenced by proteomic and microscopic analyses. The treatment of this bacterium could prove challenging due to its high-level synthesis of GES-5 β-lactamase conferring resistance to last resort antibiotics like carbapenems. A possible alternative could be the usage of lytic bacteriophages, as they lyse their host very specifically (Strathdee et al., 2023). The identified lytic bacteriophage vB_CbrP_HGW_001 is able to effectively lyse GW-Imi-1b1. It is noteworthy that the similarity in GC content between the phage and its host strain may reflect their close co-evolution. Although GC contents of eukaryotic viruses do not appear to be associated with the ones of their hosts, comparable GC contents and codon usage patterns between phages and their respective prokaryotic hosts have previously been reported (Simón et al., 2021). These similarities may therefore reflect the close adaptation of phages to their bacterial hosts and are consistent with the narrow host specificity observed for vB_CbrP_HGW_001. On the one hand, a high host specificity is advantageous because it ensures that no other species are affected by the phage. Conversely, a narrow host range generally restricts treatment options, as a monophage therapy is not effective against polymicrobial infections and can result in mismatches between the phage and various strains of the respective pathogen. The utilization of phage cocktails, a combination of diverse lytic phages, can circumvent this limitation, yielding a more extensive spectrum of activity against pathogens, albeit still being more selective compared to antibiotics with a narrow spectrum (Loc-Carrillo and Abedon, 2011; Oromí-Bosch et al., 2023). Notably, the lytic host range differs in the genus Kayfunavirus, as both broad host range phages and phages infecting only their propagation strain have been reported (Sun et al., 2022; Liu et al., 2024; Mutai et al., 2026). The here described phage lyzes only its propagation strain. However, a broader test of the host spectrum would be necessary in order to precisely determine the host range.

The average burst size within the genus Kayfunavirus typically ranges from 35–55 PFU/cell (Sun et al., 2022; Cao et al., 2023; Vidigal and Hungaro, 2024). In contrast, vB_CbrP_HGW_001 exhibited a much smaller burst size of 12 PFU/cell. However, other phages with podovirus morphology, such as coliphage vB_Eco_P_SU7, have been reported to display comparable burst sizes (Koonjan et al., 2021). Notably, small burst sizes are often associated with short lysis times, thereby enabling earlier infection of other host cells (Shao and Wang, 2008), which is consistent with the adsorption assay results obtained in this study. In comparison, previously described phages with a higher burst size exhibit longer latent periods of 9–15 min (Sun et al., 2022; Cao et al., 2023; Vidigal and Hungaro, 2024), whereas vB_CbrP_HGW_001 showed a latent period of approximately 5 min. Thus, the short latent period of vB_CbrP_HGW_001 may serve to counterbalance its comparatively diminished burst size. In clinical settings, the reduced burst size could potentially be compensated by higher initial dosing, as the latent time may contribute to limiting the emergence of phage-resistant bacterial variants. Notably, all assays conducted in this study were performed in vitro. In vivo properties of vB_CbrP_HGW_001 could differ from presented results. Further preclinical trials are required to assess suitability of vB_CbrP_HGW_001 for phage therapy, as testing its lytic ability ex vivo and in vivo, as well as its potential cytotoxicity toward eukaryotic cell lines (Nale and Clokie, 2021).

A study by Liu et al. (Liu et al., 2024) reported a lytic bacteriophage belonging to the Myoviridae against another C. braakii strain. Similar to vB_CbrP_HGW_001 in this study, capability of infection was affected in basic and acidic pH value and at temperatures higher than 50 °C. Furthermore, they exhibited synergistic effects of the bacteriophage with different antibiotics in vitro (Liu et al., 2024). As bacteria can escape phage infection through various mechanisms, e.g., altering surface phage receptors to prevent phage adsorption or defense systems such as CRISPR or restriction-modification (R-M) systems (Oromí-Bosch et al., 2023), it is crucial to develop treatment strategies that impede phage resistance development. For the treatment of enterococcal infections, it has been reported that a combination of phage therapy and systemic antibiotics application can lead to fewer hospitalizations and to a reduction of the pathogen incidence in the patient’s stool (Stellfox et al., 2024). Future studies could thus focus on generally combining bacteriophages with antibiotics to treat problematic infections caused by multi-resistant bacteria.

In this study, a Citrobacter braakii strain was isolated from hospital sewage in Greifswald, Germany. The multiresistant strain was further characterized and revealed a constitutively expressed GES-5 β-lactamase, which was found to be among the 10 most abundant proteins in the cellular proteome. This protein is predicted to confer resistance to carbapenems which may complicate treatment of such bacterial infections.

From the sewage, a lytic bacteriophage for the Citrobacter strain was successfully isolated and characterized in detail. The phage represents a promising alternative treatment option against GW-Imi-1b1. Further research is required to investigate a putative increased efficacy of combining this phage with antibiotics in a possible patient treatment scenario. The presented workflow is suitable for the isolation and characterization of bacteriophages from sewage targeting clinically relevant and most contemporary pathogens. Furthermore, global proteomics proved to be a powerful tool to reveal actively expressed genes including virulence factors and resistance mechanisms of problematic pathogens paving the way for novel and targeted therapies.